Geophys. J . In(. (1992) 109, 162-170 Correspondence between theory and observations of polar motion Richard S. Gross Jet Propulsion Laboratory, California Institute of Technology, Pasadena, California 91 109 USA Accepted 1991 November 4. Received 1991 November 4; in original form 1991 March 29 SUMMARY The Earth’s orientation in space changes in response to the action of a variety of torques, generated both externally and internally. External torques arising from the gravitational forces of the Sun and Moon act upon the Earth, causing it to undergo periodic motions known as the lunisolar precession and nutations. Internal dynamical processes that change the deformable Earth’s inertia tensor, or that generate relative angular momentum, also cause the Earth’s orientation in space to change, the resulting motions being known as wobble (or polar motion) and changes to the length-of-day. Historically, it was thought that observations determined the location of the rotation pole within some rotating reference frame fixed to the solid Earth, and theoretical expressions for the nutations and wobble were developed and given in terms of the Earth’s rotation axis. However, it has been (relatively recently) argued that observatories, being located on the surface of the Earth, are more nearly moving with the Earth’s surface, and hence observations more nearly reflect the motion of the Earth’s figure axis, rather than its rotation or angular momentum axes. This argument was accepted by the International Astronomical Union, and the current 1980 Theory of Nutation refers to an axis, the celestial ephemeris pole, that is more closely associated with the Earth’s figure axis than it is with the Earth’s rotation axis. Polar motion values, as reported by modern Earth rotation services, give the location of the celestial ephemeris pole within some rotating, body-fixed reference frame. The celestial ephemeris pole does not correspond to either the Earth’s instantaneous figure axis, its instantaneous rotation axis, or its instantaneous angular momentum axis, but rather corresponds to an axis that exhibits no nearly diurnal motions in either the terrestrial, body-fixed reference frame or the celestial, space-fixed frame. The focus of this paper is not on the nutational motions of the Earth generated by external torques, but rather on the wobble. The goal of this paper is to write in terms of reported values the standard theoretical equation describing the Earth’s wobble, namely, the linearized conservation of angular momentum equation known as the Liouville equation. In terms of the location m ( t ) = r n , ( t ) + i r n 2 ( t ) of the rotation pole, the Liouville equation is usually written as m i dm + -=~ CJO dt i d ( t-) - - x ( t ) !2 dt where the complex-valued X-functions are functions of perturbations to the Earth’s inertia tensor and relative angular momentum. In order to rewrite this equation in terms of the location p ( t ) = p l ( t ) ipz(t) of the reported pole, a relation between the locations of the rotation pole and the celestial ephemeris pole must be obtained. This is accomplished in the time domain by considering the properties of the time-dependent transformation matrix that relates components of a position vector in the rotating, body-fixed frame to its components in the celestial, space-tixed frame, the resulting relation being + i dp m ( t ) = p ( t )- --. 52 dt 162 Theory and observations of polar motion 163 Using this expression relating the location of t h e rotation pole to that of t h e celestial ephemeris pole, the linearized conservation of angular momentum equation becomes Thus, there is no 2-term in t h e Liouville equation describing long-period polar motions when it is written in terms of reported valves. Key words: E a r t h rotation, length-of-day, polar motion, wobble. INTRODUCTION Conceptually, the Earth’s rotation is easily understood. External torques operate on the Earth, affecting the Earth’s angular momentum vector, causing it to change in space. External is taken here to mean everywhere external to the surface of the Earth, with the Earth’s atmosphere and hydrosphere being considered to be internal to the Earth’s surface. By definition, the angular momentum vector is the inner product of the Earth’s inertia tensor with its rotation vector. Thus, changes of the Earth’s inertia tensor and rotation vector will be associated with changes of the angular momentum vector. The resulting (time-dependent) changes of the Earth’s orientation in space generated by these external torques are known as the luni-solar precession and nutations. Internal processes can also change the Earth’s inertia tensor, or generate relative angular momentum, thereby changing the Earth’s rotation. However, in the case of internal processes, the Earth’s total angular momentum vector is conserved and does not change in space. Complications arise in practice due to the Earth’s ability to deform in response to forces of both external and internal origin, causing the Earth’s inertia tensor to change, thereby also changing the Earth’s rotational motion. In the past, there has been some confusion concerning the observations of the rotational motions of the Earth, much of it centred on understanding what is actually being determined by these observations. Is it the location of the Earth’s rotation axis, angular momentum axis, figure axis, or some other axis? Historically, it was thought that observations determine the location of the Earth’s rotation axis. Theories of the Earth’s rotation, encompassing both the nutations generated by the gravitational forces of heavenly bodies external to the Earth (e.g. Woolard 1953), and the wobble and changes in the length-of-day generated by forces internal to the Earth (e.g. Munk & MacDonald 1960), were developed and expressed in terms of the Earth’s rotation axis. It was only relatively recently that Jeffreys (1959, 1963) and, later, Atkinson (1973) argued that observatories, being located on the surface of the Earth, are moving in space more nearly with the Earth’s figure axis, rather than its rotation or angular momentum axes, and that, therefore, a theory of nutation should refer to the figure axis. This point of view was accepted by the International Astronomical Union (IAU), and their 1980 Theory of Nutation refers to an axis associated with the Earth’s figure axis (Seidelman 1982). The subject of this paper, however, is not the nutational motion of the Earth generated by external torques, but rather the rotational motion of the Earth known as the wobble and changes in the length-of-day that are generated by internal dynamical processes. The theory of the wobble is currently written in terms of the location of the rotation axis. However, Earth rotation services do not report the location of the Earth’s rotation axis, but rather report the location of the celestial ephemeris pole, which is an axis more closely associated with the Earth’s figure axis than it is with the Earth’s rotation axis. Classical astronomic latitude-longitude observations, as well as observations by the modern space-geodetic techniques of very long baseline interferometry (VLBI), satellite laser ranging (SLR) and lunar laser ranging (LLR), can determine the orientation in space of a terrestrial reference frame defined by the locations of the observatories (see below). That is, Earth rotation observations can determine the orientation of a rotating, body-fixed reference frame (defined by the location of a set of observatories) with respect to a celestial, space-fixed reference frame (defined by the location of a set of stars or radio sources, or defined by the orbital motion of the Moon or some artificial satellite). The goal of this paper is to write the theory of the Earth’s rotation (describing the Earth’s wobble and changes in the length-of-day) in terms of reported values. It will be shown that the difference between the linearized conservation of angular momentum equation written in terms of the location of the rotation axis (which is its usual form) and the Liouville equation derived in this paper (written in terms of the location of the celestial ephemeris pole) is only important for the wobble, with no modifications being needed to the equation describing changes in the rotation rate (or length-of-day) of the Earth. Observations can determine the Earth’s orientation in space, but cannot distinguish between an external or internal origin for some observed change in the Earth’s orientation. For simplicity, it will be assumed throughout this paper that no external torques are acting upon the Earth model. In this case, changes in the Earth’s rotation are due solely to internal dynamical processes, the resulting motions being known as wobble, or polar motion, and changes in the length-of-day. THEORY OF THE EARTH’S ROTATION Consider some deformable Earth model that is rotating with respect to an inertial reference frame (the i-frame). Let this Earth model be observed from a reference frame that is itself rotating with angular velocity w with respect to the inertial reference frame. This rotating reference frame will later be attached to the Earth model in some prescribed manner, but for now consider w to be completely arbitrary. R . S. Gross 164 Conservation of angular momentum within the rotating reference frame is expressed by: their time derivatives (Wahr 1982): dL =Q2(C - A ) -twXL=t (Q2 Af,, + 9 A h , + Q- dt at (9) where the time derivative denotes change with respect to the rotating reference frame, and t represents external torques acting o n the Earth model. T h e angular momentum L can in general b e separated into two terms as + L=I.w h (2) where I is the inertia tensor, and h is the relative angular momentum due t o motion of the Earth model relative to the rotating reference frame. Initially, let the Earth model be in a state of uniform rotation at the rate 51 about its figure axis which coincides with the &-axis of the inertial reference frame (the gi-axis). Let the rotating reference frame be attached to the Earth model such that it is also initially rotating at the rate Q about its %-axis which coincides with the i$-axis. Furthermore, let the rotating reference frame be oriented within the Earth model such that the inertia tensor of the Earth model is diagonal within the rotating reference frame. Finally, let no external torque t or relative angular momentum h be present: where C,,, is the greatest principal moment of inertia of the mantle and the factor 1.61 accounts for the nonparticipation of the core in the wobble and the assumption that the solid Earth and oceans respond linearly to changes in rotation. By defining the complex quantities m = m 1 + im,, Ah = A h , q = ql + iy,, + i Ah,, Al= At,, (12) + i At,,, equations ( 6 ) and (7) can b e written more compactly as i m+-m=q 0, where by (9) and (10): ' = Q2(C- A ) where t h e subscript o denotes initial quantities. This initial state is perturbed by some internal geophysical event causing (in general time dependent) perturbations A1 to the inertia tensor and Ah to the relative angular momentum, which, by conservation of angular momentum, induces perturbations Am to the angular velocity. The perturbations to the angular velocity a r e usually written in terms of the dimensionless quantities mi defined by Am = Q ( m , , m,, m,)*. (4) Assuming that all perturbations are small, Im,J << 1, 1AJJ << C, JAhil<< QC, (5) so that their products can be neglected, it can be shown (e.g. Munk & MacDonald 1960; Lambeck 1980; Wahr 1982; Barnes et al. 1983) that the linearized conservation of angular momentum equation is 1 ml--m2=vl, 00 1 m,+-m, 0" = v3, + Q A h - iQ __ d' dt - i-d A h ) dt . (14) The total rotation vector is w = W, + AW (15) which by (3) and (4) has components in the rotating reference frame: + w = Q ( m , , m,, 1 m3)T. ( 16) Since, to first order in the m i , the magnitude lo(of the rotation vector is Q(l + m,), then m i , m,, 1 are the direction cosines of the rotation vector with respect to the coordinate axes of the rotating reference frame. Alternatively, and also t o first order in the mi, m l and m , give the angular offset of the rotation vector from the %-axisof the rotating reference frame in the t- and ?-directions, respectively, of the rotating reference frame. If the Earth model were rigid, then all its mass elements would have the same angular velocity with respect to the inertial reference frame, and the rotating reference frame could simply be chosen t o rotate with this angular velocity. In this case, the velocity in space of some mass element of the Earth model located at r is simply given by dr _ -oXr. = q2, (7) (8) where the dot denotes time differentiation, and uo is the complex-valued frequency of the Chandler wobble. T h e vi are known as excitation functions which, in the absence of external torques, are functions of the perturbations t o the inertia tensor and the relative angular momentum as well as m3 (SZ2 A 1 dt However, since the Earth model is deformable, different mass elements may, in general, rotate with different angular velocities. In this case, the rotating reference frame can be chosen to rotate with some mean angular velocity ( w ) defined such that the quantity Theory and observations of polar motion is a minimum where p is the density of the mass element located at r and the volume integral extends over that part of the Earth model of interest. If this volume integral extends over the crust and mantle of the Earth model, then the rotation vector thus defined determines the ‘Tisserand mean mantle’ rotation vector (e.g. Munk & MacDonald 1960, p. lo), and it can be shown (e.g. Jeffreys 1976, p. 293) that within a reference frame rotating with the ‘Tisserand mean mantle’, the relative angular momentum of the crust and mantle is zero. Wahr (1981) has generalized the concept of the ‘Tisserand mean mantle’ to that of the ‘Tisserand mean surface’ by replacing the volume integral in (18) with a surface integral extending over the surface of the Earth model. Since the stations that observe the Earth’s rotational properties are located on the surface of the solid Earth, observations of the Earth’s rotational properties most nearly reflect the rotation of the ‘Tisserand mean surface’. The rotating reference frame defined herein is chosen to rotate with the ‘Tisserand mean surface’, and is thereby considered to be ‘fixed’ to the Earth model. At this point the motion of the rotating, body-fixed reference frame has been specified. It now remains to specify its orientation within the Earth model. The origin of the rotating, body-fixed reference frame is first placed at the centre-of-mass of the Earth model. The terrestrial frame, or t-frame, is then defined by aligning its coordinate axes with the axes of some chosen conventional terrestrial reference frame (CTRF). A CTRF is defined in practice by specifying the coordinates of a set of observatories located on the surface of the Earth. Examples of conventional terrestrial reference frames are the Conventional International Origin (CIO; e.g. Lambeck 1988, p. 27) and the various solutions of the International Earth Rotation Service (IERS) Terrestrial Reference Frame such as ITRF89 (see the IERS Annual Report for 1989). The results of Earth rotation observations are usually given within some CTRF. O B S E R V A T I O N S OF T H E E A R T H ’ S ROTATION The theory of the Earth’s rotation, as outlined above, relates changes of the rotation vector to the geophysical processes causing these changes. Changes of the rotation vector are specified by the values of the mi which locate the rotation axis within a rotating, body-fixed reference frame. If equations (6-8) are to be used to infer properties of the Earth from Earth rotation observations, then these observations must yield the time-dependent location of the rotation axis within this rotating, body-fixed reference frame. Historically, this was thought to be the case, and for this reason Woolard (1953) developed a series of coefficients describing the nutational motion of the rotation axis of a rigid Earth model. These coefficients determined by Woolard (1953) were then adopted by the International Astronomical Union (IAU) as part of their System of Astronomical Constants for use starting in 1960. However, currently operating Earth rotation services do not report the location of the rotation pole within some rotating, body-fixed reference frame, but rather report the location of a pole, known as the celestial ephemeris pole, that is more closely associated with the Earth’s figure axis than it is with the Earth’s rotation axis. 165 Jeffreys (1959, 1963) and Atkinson (1973) argued that since observatories are attached to the surface of the solid Earth, and that since it is the nutational motion of the observatories that is actually observed, then for theory to correspond to observations, the theoretical series of nutational coefficients should be given for the figure axis of some Earth model. In the case of a deformable Earth model care must be taken in specifying this figure axis, but they argued that conceptually the figure axis should be used since this is the axis (rather than the rotation or angular momentum axes) that is most firmly ‘tied’ to the locations of the observatories. This point of view was accepted by the IAU when recommending that the series of nutation coefficients for the B-axis of Wahr (1981) be adopted as the 1980 IAU Theory of Nutation. Brief histories of these events have been given by Kinoshita et al. (1979) and Seidelmann (1982). Atkinson (1975) explicitly showed that meridian observations of stellar declinations involve the Earth’s axis of figure, concluding that nutations should be computed for this axis, rather than the axis of rotation. Ooe & Sasao (1974) have explicitly shown that astronomical latitude observations at a single station, involving measurements of the zenith distances of stars at their meridian passage [see, e.g., Munk & MacDonald (1960) for a review of the optical astrometric observing techniques], determines the orientation of the local vertical in space, and that simultaneous observations at a number of stations determines the location of the reference axis of the star catalogue within the conventional terrestrial reference frame. Similarly, Ooe & Sasao (1974) have explicitly shown that astronomical longitude observations, involving timing the meridional passage of zenith stars, also involves the location of the reference axis of the star catalogue within the conventional terrestrial reference frame. Thus, classical astronomic latitude-longitude observations at an individual station determine the orientation in space of the station’s local vertical (Jeffreys 1959, 1963; Atkinson 1973, 1975; Ooe & Sasao 1974; Kinoshita et al. 1979; Sasao & Wahr 1981; Capitaine, Williams & Seidelmann 1985), and simultaneous astronomic latitudelongitude observations at a globally distributed set of stations determine the orientation in space of some reference frame attached to the stations. Since the stations are attached to some conventional terrestrial reference frame by the values of their nominal latitudes and longitudes, then classical astronomic latitude-longitude observations determine the orientation of the t-frame with respect to the i-frame. The very long baseline interferometry (VLBI) technique measures the difference in the arrival time of a radio signal at two or more radio telescopes that are simultaneously listening to the same distant source (e.g. Shapiro 1983; Lambeck 1988, chapter 8). This technique is therefore sensitive to processes that change the relative position of the radio telescopes with respect to the source, such as a change in the orientation of the Earth in space. If just two telescopes are observing the same sources, then only two components of the Earth’s orientation in space can be determined. A change in the orientation of the Earth about an axis parallel to the baseline vector connecting the two radio telescopes does not change the relative position of the telescopes with respect to the source, and hence this 166 R . S. Gross component of the Earth’s orientation is not determinable from single baseline VLBI observations. Multibaseline VLBI observations with satisfactory geometry can determine all three components of the Earth’s orientation with respect to a reference frame determined by the locations of the radio sources. Thus, VLBI measurements can determine (among other things) the orientation of a conventional terrestrial reference frame (defined by the adopted latitudes and longitudes of a set of radio telescopes) with respect to an (herein assumed) inertial, conventional celestial reference frame (defined by the adopted declinations and right ascensions of a set of distant radio sources). In the technique of satellite laser ranging (SLR), the round-trip time of flight of laser light pulses travelling between a ground-based laser and a retroreflector mounted on some satellite is accurately measured (e.g. Lambeck 1988, chapter 6). This time-of-flight measurement is converted into a distance, or range, measurement by using the speed of light and correcting for such effects as atmospheric path delay, satellite centre-of-mass offset, etc. The SLR technique is therefore sensitive to any process that changes the distance between the retroreflector mounted on the satellite and the ground-based laser system, such as that caused by a change in the Earth’s orientation. Thus, SLR measurements can determine (among other things) the orientation of a conventional terrestrial reference frame (defined by the adopted latitudes and longitudes of a set of laser ranging stations) with respect to a reference frame determined by the satellite’s orbit. Lunar laser ranging (LLR) is a technique similar to that of satellite laser ranging but where the retroreflector is located on the surface of the Moon (placed there by Apollo astronauts and unmanned Soviet spacecraft), rather than on some satellite. LLR can therefore determine the orientation of a conventional terrestrial reference frame with respect to a reference frame determined by the Moon’s orbit. In the reduction of the observing data taken by the modern, space-geodetic observing techniques (such as VLBI), the body-fixed coordinates (after accounting for the solid-Earth and ocean-loading tidal displacements) of some observing station are transformed to its coordinates in an inertial (or quasi-inertial) reference frame by applying a series of rotation matrices. Let r,(t) denote the components of some position vector as given in some rotating, body-fixed reference frame (this position vector may be time-dependent due to, for instance, plate-tectonic motions). Let ri(t) denote the components of this same position vector but now given in some inertial, space-fixed reference frame. These two components of the same position vector are related to each other by a frame transformation, which can be written as a series of rotation matrices (e.g. Sovers 1991): ri(t) = PNUXYr,(t). (19) where P accounts for the precession of the Earth, N accounts for the nutational motion of the Earth, U accounts for the spin of the Earth, and the product XY accounts for the Earth’s polar motion. Models (or pre-determined values) of the Earth’s precession, nutation, spin, and polar motion are used to fill the elements of the P, N, U, X, and Y matrices when actually applying (19) to transform the body-fixed coordinates of some observing station to its space-fixed coordinates. Observations are then fit to computed values in order to obtain corrections to the a priori model parameters. The currently accepted model for the Earth’s precession and nutations (Seidelmann 1982; McCarthy 1989) applies to the celestial ephemeris pole, which coincides with the B-axis of Wahr (1981). The B-axis of Wahr (1981) is an averaged figure axis of Earth model 1066A (Gilbert & Dziewonski 1975) where the averaging procedure is meant to average out the effects of the body tides on the figure axis. Thus, the celestial ephemeris pole has the desired property of not being displaced in response to the body tides, where the instantaneous figure axis does move in response to the body tides. Following the precession P and nutation N matrices, the next rotation matrix in the sequence of transformations (19) is the matrix U which represents a large rotation about some axis. The axis about which this large rotation is taken is the celestial ephemeris pole, since this is the axis to which the precession P and nutation N rotation matrices apply. Similarly, the product of the transformation matrices X and Y represents the polar motion of the celestial ephemeris pole. That is, the elements of the rotation matrices X and Y involve the location of the celestial ephemeris pole within the rotating, body-fixed reference frame (e.g. Sovers 1991): XY=( cosPMX 0 -sinPMX 0 1 sinPMX 0 cosPMX X =( i‘ 0 0 COSPMY sinPMY 0 -sinPMY cosPMY O ) z;) O1 01 (20) PMX -PMY 1 where PMX and PMY are the polar motion quantities reported by Earth rotation services (with PMY being positive towards 90“W longitude), and where series expansions for the sine and cosine functions have been used in order to obtain the final matrix, with terms of second order and higher in the small quantities PMX and PMY being neglected. In the above, the celestial ephemeris pole was identified with the B-axis of Wahr (1981). This was done because the elements of the nutation matrix N are filled with the results of a model for the nutational motions of Wahr’s B-axis. Each of the transformations P, N, U, X, and Y must refer to the same axis. Thus, if one transformation matrix explicitly refers to the B-axis, then they must all refer to this same axis. However, Moritz & Mueller (1988, section 3.6) argue that the celestial ephemeris pole should be interpreted as an averaged angular momentum axis of the Earth. For the purposes of this paper, the physical interpretation of the celestial ephemeris pole is unimportant. It is merely sufficient to realize that the celestial ephemeris pole corresponds to that axis about which the large rotation is taken in the transformation matrix U . It is this realization that is exploited below when deriving an expression for the location of the rotation axis in terms of the location of the celestial ephemeris pole. Theory and observations of polar motion Thus, to summarize this section, modern Earth rotation services do not report the location of the rotation axis within some rotating, body-fixed reference frame [that is, Earth rotation services do not directly report the m i that are needed in equations (6-8)], but rather report the location of the celestial ephemeris pole. This does not imply that the modern, space-geodetic observing techniques are not sensitive to the location of the Earth’s instantaneous rotation vector. It is merely an accepted convention that the location of the celestial ephemeris pole be reported, rather than the location of some other axis such as the Earth’s instantaneous rotation axis, or its instantaneous angular momentum axis, or any other well-defined axis. The advantage of reporting the location of the celestial ephemeris pole is that this pole has the desired property of exhibiting no nearly diurnal motions in either the rotating, body-fixed reference frame or the inertial, space-fixed reference frame (Seidelmann 1982; Moritz & Mueller 1988, chapter 9). In order to relate the theory of the Earth’s rotation to observations, it is necessary to obtain an expression for the location (given within the t-frame) of the rotation axis in terms of reported quantities, namely, in terms of the location (also given in the t-frame) of the celestial ephemeris pole (which is just the axis about which the large rotation is taken in the transformation matrix U). This can be accomplished by considering the properties of the time-dependent transformation matrix that relates the components of a vector in the i-frame to its components in the t-frame. In this consideration it will be assumed that no external torques exist, and hence that the Earth model does not experience luni-solar precession and nutation. This is equivalent to assuming that the precession and nutational motions of the Earth are completely understood and can be perfectly modelled and removed from Earth rotation observations. The deformable Earth model does, however, experience wobble and variations in rotation rate. The derivation given here of the rotation vector in terms of the coordinate transformation matrix and its time rate-of-change closely follows that of Chandrasekhar (1969, section 25); also see Kinoshita et al. (1979) and Moritz & Mueller (1988, section 2.3.1). KINEMATIC DESCRIPTION OF T H E EARTH’S ROTATION Consider a body-fixed reference frame (the t-frame) rotating with respect to an inertial, space-fixed reference frame (the i-frame). Let the origins of these two reference frames coincide, but make no assumptions about their relative orientation. In particular, it is not necessary to assume that the axes of the two reference frames are momentarily coincident. There exists a transformation matrix A(t) that relates the components r,(t) of a position vector given in the inertial frame to its components rt(t) given in the body-fixed frame rI(t) = A(r)ri(t). The transformation matrix is time dependent since the t-frame is assumed to be rotating with respect to the i-frame, and hence its orientation in space is continually changing. Since A(t) represents an orthogonal transformation (e.g. Goldstein 1950, section 4.2), the inverse of A(?)is given by 167 its transpose: A(t)AT(t)= I where the superscript T denotes the transpose and I is the identity matrix. From the time derivative of (22): dA -AT(?) dr dAT + A(t) =0 dt it is clear that the matrix W ( t ) defined by dA W(t) -AT(t) dt is antisymmetric since combining (23) and (24) yields W(t) + WT(r) = 0. (25) A vector can be associated with any antisymmetric matrix, and vice versa: WJ= &*lkWk (26) where summation of repeated indices is assumed, and E , , ~is the Levi-Civita, or alternating, tensor defined to be 0 if any two of the indices ijk are identical, +1 if ijk is an even permutation of 123, and -1 if ijk is an odd permutation of 123. The elements W k of the vector o are the three independent elements of the antisymmetric matrix W. In order to assess the physical significance of w, consider the coordinate transformation of the time derivative of some vector r(t). Equation (21) relates the components ri(t) of r(t) in the inertial frame to its components rt(t) in the body-fixed frame. Inverting (21) yields r,(t) = AT(t)rl(t) (27) the time derivative of which is dri dAT dr, rI+AT-. dt dr dt Multiplying (28) by A(t) and using (21-25) produces A d r. dt d = - (Ari)- W(Ari). dt By using (26) and remembering that ( w X rJi = &j,kO,(rt)k this becomes the more familiar A dr. d = - (Aq) + w X (Ari). dt dr The left-hand side of this expression represents the time rate of change of the vector in the inertial frame, but with its components being given along coordinate axes that momentarily coincide with the coordinate axes of the body-fixed frame. The first term on the right-hand side of (30) represents the time rate of change of the vector in the body-fixed frame, and the second term on the right-hand side represents the change of the vector due to rotation of the body-fixed frame with respect to the inertial frame (e.g. Goldstein 1950, section 4.8). Thus the vector o associated with the antisymmetric tensor W defined by (24) is just the rotation vector describing the rotation of the body-fixed frame with respect to the inertial, space-fixed frame. The magnitude of o is the instantaneous angular velocity of the 168 R. S. Gross t-frame (with respect to the i-frame) as it instantaneously rotates about an axis given by the direction of w . From the definition (24) of W ( t ) , an expression for the components wi of the rotation vector in terms of the elements of the coordinate transformation matrix A ( t ) can be obtained. The derivation given here follows that of Ooe & Sasao (1974); also see Capitaine (1986a,b) and Brzezinski & Capitaine (1991). For the Earth rotation problems of interest in this paper, it is assumed that the rotation of the Earth departs only slightly from a state of uniform rotation about some axis. In order to derive expressions for the elements of A ( t ) under this assumption, it is convenient to introduce another reference frame, the u-frame, that is uniformly rotating at the rate 51 about its reference axis (i.e., the %,-axis). Let the origin of the u-frame coincide with that of the i-frame and the t-frame, and furthermore, let the reference axis of the u-frame coincide with that of the i-frame. That is, let the $-axis of the uniformly rotating reference frame (about which it is rotating) coincide with the $-axis of the inertial reference frame. A transformation matrix U ( t ) exists that relates the components ri(t) of some position vector given in the inertial frame to its components ru(t) given in the uniformly rotating frame: ru(t) = U(t)ri(t) where (e.g. Goldstein 1950, p. 109) U(t)= ( cosQt -s$Rt :). sinQt 0 coi51t In the absence of any perturbing forces, let the initial dynamical state of the Earth be that of uniform rotation at the rate 51 about the reference axis of the body-fixed frame (i.e., the %,-axis). That is, in the initial state, the t-frame coincides with the u-frame, both of which are uniformly rotating at the rate Q about the reference axis of the i-frame (which, of course, coincides with the reference axes of the t-frame and the u-frame). Under the action of some (in general, time-dependent) perturbing force, the t-frame will begin to rotate (with respect to the i-frame) about some axis (the rotation axis) that will no longer coincide with either the &-axis, the %,-axis, or the &-axis. Assume that the perturbing force is small enough that (at any time) the departure of the instantaneous rotation vector from its initial state is small. Thus, since the t-frame and u-frame initially coincided, assume that they always remain close to each other under the action of the perturbing force. In this case, at any time, the coordinate transformation matrix B(t) relating the components r,(t) of some position vector given in the uniformly rotating frame to its components r,(t) given in the body-fixed frame, (33) represents an infinitesimal transformation matrix, the general form of which is (e.g. Goldstein 1950, section 4.7) r,(t) = B ( M t ) > B(t) = 1 -P3(t) -p&) ( P3W 1 -pzW PlQ) Pdt) 1 ) paragraph, a sign convention has been used that is different from that usually employed. The significance of the elements pi of the infinitesimal transformation matrix B(c) can be understood by considering the components in the body-fixed frame of a unit vector aligned with the reference axis of the uniformly rotating frame. That is, let r,(t) = (0, 0, l)T. Then by (33) and (34), r,(t) = [ p l ( t ) ,p z ( t ) , I]*. Since r,(t) is a unit vector (to first order in the pi),then p l ( t ) , p 2 ( t ) , 1 are the direction cosines of the reference axis of the u-frame in the body-fixed frame. Or, since the reference axes of the uniformly rotating and the inertial frames coincide, p l ( t ) , p 2 ( t ) , 1 are the direction cosines of the %,-axiswith respect to the coordinate axes of the body-fixed frame. Thus, p l ( t ) and p 2 ( t ) are simply the polar motion quantities reported by Earth rotation services [compare (34) with the transpose of (20)], and, as shown below, p 3 ( t ) is related to changes in the rotation rate of the Earth model. Note that in the absence of external torques, so that the Earth model does not undergo h i - s o l a r precession or nutation, then the total angular momentum vector of the Earth is constant in space, and hence its direction can be chosen to coincide with that of the reference axis of the inertial frame. Thus, in the absence of external torques, p l ( t ) and p 2 ( t ) locate the Earth’s total angular momentum vector in the body-fixed reference frame. But, more generally, p , ( t ) and p 2 ( t ) locate the reference axis of the u-frame (and of the i-frame since these two reference axes coincide) in the t-frame. In order to determine the components w, of the rotation vector in the body-fixed reference frame, combining (31) and (33) yields r,(t) = B(t)U(t)ri(t), so that by (21), A ( t ) = B ( t ) U ( t ) , and thus by (24), W ( t ) becomes (to first order in the pi): dP,+ 51 dt dt 0 ---p251 ddtp l *,P,Q dt \ I . (35) By correspondence with (26), the components of the rotation vector in the body-fixed reference frame are then w,(t) = QP,(t) +!j2, w2@)= QPZ(t) - dl? (36) (37) d t ) = Q +A, (38) where the dot denotes time differentiation, and hence by (16), the m iin terms of the reported valves pi are (39) (34) where here, for reasons discussed below in the next Equations (39-41) are the desired expressions giving the location of the rotation vector in the rotating, body-fixed Theory and observations of polar motion reference frame in terms of the location of the celestial ephemeris pole. In order to interpret (41), note that m3 is related to the time interval (UT1 - TAI) by d m,(t)=-(UTl dt - TAI) (42) where UT1 represents a time-scale kept by the rotating Earth, and TAI is a reference time-scale based upon atomic clocks (e.g. Lambeck 1980, p. 63). By combining (41) and (42) it is therefore seen that p 3 ( t ) , apart from a constant of integration, is the angle about the ?,-axis (or the %,-axis since they coincide) through which the Earth has rotated during the time interval (UTl - TAI): p 3 ( t ) = Q(UT1 - TAI). (43) Of more interest here are the results given by (39) and (40).By defining p ( t ) as P(t) = d t ) + ipz(t) (44) and using (39) and (40), then (13) becomes By defining x ( t ) from it is clear that (45) is satisfied by p(t) + L ~ ( t=)x ( t ) . (47) *o Expression (47) is one of the main results of this paper. It relates reported values of polar motion to the geophysical mechanisms causing the Earth’s rotational properties to change. The reported values p ( t ) are the location within the body-fixed reference frame of the reference axis of the uniformly rotating reference frame (which coincides with the reference axis of the inertial, space-fixed reference frame). By (14) and (46), x ( t ) is a function of the perturbations to the inertia tensor and relative angular momentum: 1.61 ‘ ( ‘ ) = Q ( c - A ) [Q AZ(t) + Ah(t)]. 148) The X-functions were defined by Barnes et al. (1983) as being more accurately computable from available meteorological data that are the ap-functions. Here it has been shown that the X-functions naturally follow from the result of writing the theoretical equation describing variations of the Earth’s rotation (13) in terms of reported values. DISCUSSION Polar motion has been historically defined as the motion of the rotation axis with respect to the figure axis as observed within some rotating, body-fixed reference frame and has been extensively studied using (13) with v(t) given by (14). However, the location of the rotation axis is not actually reported by Earth rotation services, and when the linearized conservation of angular momentum equation (13) is written in terms of reported values, the resulting expression is (47) 169 with ~ ( t given ) by (48). Mathematically, (47) has the same form as (13), and hence the motion within the t-frame of the location of the reference axis of the u-frame can be expected to exhibit all of the characteristics of polar motion. The right-hand side of (47), x ( t ) , can be viewed as a forcing, or excitation, function similar to v(t).A periodic variation of ~ ( t forces ) p ( f ) to undergo an oscillation at that same period, exactly analogous to the motion of m ( t ) forced by ap(t). In the absence of any forcing, p ( t ) experiences a free wobble identical to the motion of the rotation axis known as the Chandler wobble. Expression (13) is not valid at high frequencies as it was derived under a number of long-period assumptions (e.g. Wahr 1982, 1986). Notably, long-period approximations were made to the equation of motion, in particular to the centripetal acceleration terms, and to the response of the core. Also, the oceans were assumed to remain in equilibrium during the wobble, which is probably valid at low frequencies but is unlikely to remain valid at sufficiently high frequencies. No long-period approximations have been made here in deriving (39-41 j and these relations between the location of the rotation axis and that being reported by Earth rotation services are valid at all frequencies. However, since (13) is valid only at long periods, then (47) is also valid only at long periods. Evidence for this long-period validity of (47) is readily obtained upon its Fourier transformation, yielding P(w ) = H ( w ) X (o) where P ( w ) is the Fourier transform of p ( t ) , X ( w ) is the Fourier transform of ~ ( t ) and , H ( w ) = u,(u, - w ) - ’ is known as the Earth’s transfer function. This transfer function is only resonant at the Chandler frequency a,, and not at the free core nutation frequency. In order for any transfer function to be valid at high, nearly diurnal frequencies, it must exhibit a resonance at the free core nutation frequency. The difference between the location of the Earth’s rotation axis and that reported by Earth rotation services is only important at high (nearly diurnal) frequencies, Using (12) and (44)to rewrite the equations (39) and (40) giving the location of the rotation axis m ( t ) in terms of the reported values p ( t ) produces i dp m ( t ) = p ( t )- -52 d f (49) which, upon Fourier transformation, becomes M ( w ) = (1 +X)P(w) in the frequency domain, where M ( w ) is the Fourier transform of m(t). At frequencies w < < Q , the term in parentheses in (50) becomes nearly 1 and the motion of the reported axis p ( t ) is nearly identical to that of the rotation axis m ( f ) . At higher, closer to diurnal, frequencies, however, the term in parenthesis in (SO) is no longer near 1 and the reported axis can no longer be identified with the rotation axis. Expressions (47) and (49) are the main results of this paper. Expression (49), valid at all frequencies, relates the location of the rotation axis to that actually reported by Earth rotation services. Expression (47), valid at long periods, is the Liouville equation written in terms of reported values. In the past, results being reported by Earth 110 R . S. Gross rotation services have been interpreted using expression (13), with q(t) being given by (46). This differs from the present result (47) by the absence [in (47)] of the i - t e r m . Eubanks et al. (1990) and Eubanks (1991) have qualitatively argued on the grounds of the separability of nutations from high-frequency wobbles that the i - t e r m should be dropped when interpreting results of Earth rotation observations. Quantitative justification for the neglect of the i - t e r m when interpreting long-period polar motion results reported by Earth rotation services has been given in this paper. One of the main results of this paper, expression (47), was derived under the simplification that no external torques are acting upon the Earth model. Thus any change in the rotational motion of the Earth model in space is assumed t o be due solely t o internal dynamical processes. However, in the presence of external torques, the Earth model will experience additional rotational perturbations known as the luni-solar precession and nutations. Observations are only able to determine the orientation of a rotating, body-fixed reference frame in space and cannot determine whether a change in this orientation is due t o an external torque or due to some internal dynamical process. A spatially long-period change in the orientation of the Earth becomes nearly diurnal when viewed within a rotating, body-fixed frame. Thus, observationally, the effect of some internal dynamical process acting a t a nearly diurnal frequency cannot be separated from a n externally generated nutation. In other words, in the presence of external torques, the question of how to separate nearly diurnal wobbles from nutations must be addressed. It is beyond the scope of the present paper t o discuss this issue, but it will be addressed in a future paper. ACKNOWLEDGMENTS Discussions with J. A. Steppe, and the comments of one of the referees, T. A. Herring, led t o substantial improvement of this manuscript. The work described in this paper was performed at t h e Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. REFERENCES Atkinson, R. d’E., 1973. On the “dynamical variations” of latitude and time, Astr. J., 78, 147-151. Atkinson, R. d’E., 1975. On the Earth’s axes of rotation and figure, Mon. Not. R. astr. SOC, 171, 381-386. Barnes, R. T. H., Hide, R., White, A. A. & Wilson, C. A., 1983. Atmospheric angular momentum fluctuations, length-of-day changes and polar motion, Proc. R. SOC. Lond., A, 387, 31-73. Brzezinski, A. & Capitaine, N., 1991. The use of the precise observations of the celestial ephemeris pole in the analysis of geophysical excitation of Earth rotation, paper presented at the XXth General Assembly of the International Union of Geodesy and Geophysics, August 11-24, Vienna. Capitaine, N., 1986a. The Earth rotation parameters: conceptual and conventional definitions, Astr. Astrophys., 162, 323-329. Capitaine, N., 1986b. The conceptual and conventional definitions of the Earth rotation parameters, in Earth Rotation: Solved and Unsolved Problems, pp. 9-24, ed. Cazenave, A., Reidel, Dordrecht. Capitaine, N., Williams, J . G. & Seidelmann, P. K., 1985. Clarifications concerning the definition and determination of the celestial ephemeris pole, Asfr. Astrophys., 146, 381-383. Chandrasekhar, S., 1969. Ellipsoidal Figures of Equilibrium, Dover, New York. Eubanks, T. M., 1991. Fluid normal modes and rapid variations in the rotation of the Earth, paper presented at the XXth General Assembly of the International Union of Geodesy and Geophysics, August 11-24, Vienna. Eubanks, T. M., Carter, M. S., Josties, F. J., Matsakis, D. N. & McCarthy, D. D., 1990. Rapid changes in polar motion, paper presented at the American Geophysical Union Fall Meeting, December 3-7, San Francisco. Gilbert, F. & Dziewonski, A. M., 1975. An application of normal mode theory to the retrieval of structural parameters and source mechanisms from seismic spectra, Phil. Tram. R. SOC. Lond., A, 278, 187-269. Goldstein, H., 1950. Classical Mechanics, Addison-Wesley, Reading, MA. Jeffreys, H., 1959. Nutation: Comparison of theory and observations, Mon. Not. R. astr. SOC.,119, 75-80. Jeffreys, H., 1963. Foreword, in Nutafion and Forced Motion of the Earth’s Pole, pp. vii-xix, ed. Fedorov, Ye. P., Pergamon, Oxford. Jeffreys, H., 1976. The Earth: Its Origin, History and Physical Constitution, 6th edn, Cambridge University Press, New York. Kinoshita, H., Nakajima, K., Kubo, Y., Nakagawa, I., Sasao, T. & Yokoyama, K., 1979. Note on nutation in ephemerides, Publs Int. Latit. Obs. Mizusawa, 12, 71-108. Lambeck, K., 1980. The Earth’s Variable Rofation: Geophysical Causes and Consequences, Cambridge University Press, New York. Lambeck, K., 1988. Geophysical Geodesy: The Slow Deformations of fhe Earth, Oxford University Press, New York. McCarthy, D. D. (ed.), 1989. IERS Standards (1989), IERS Technical Note 3, Central Bureau of IERS, Observatorie de Paris. Moritz, H. & Mueller, I. I., 1988. Earth Rotation: Theory and Observation, Ungar, New York. Munk, W. H. & MacDonald, G. J. F., 1964. The Rotation of the Earth: A Geophysical Discussion, Cambridge University Press, New York. Ooe, M. & Sasao, T., 1974. Polar wobble, sway and astronomical latitude-longitude observations, Publs Int. Latit. Obs. Mizusawa, 9, 223-233. Sasao, T. & Wahr, J . M., 1981. An excitation mechanism for the free ‘core nutation’, Geophys. J. R. astr. SOC., 64, 729-746. Seidelmann, P. K., 1982. 1980 IAU theory of nutation: The final report of the IAU working group on nutation, Celest. Mech., 27, 79-106. Shapiro, 1. I., 1983. Use of space techniques for geodesy, in Earthquakes : Observation, Theory and Interpretation, pp. 530-568, eds Kanamori, H. & Boschi, E., North Holland, Amsterdam. Sovers, 0. J., 1991. Observation model and parameter partials for the JPL VLBI parameter estimation software “MODEST”1991, JPL Publication 83-39, Rev. 4, Jet Propulsion Laboratory, Pasadena, CA. Wahr, J. M., 1981. The forced nutations of an elliptical, rotating, elastic and oceanless Earth, Geophys. J. R. astr. SOC., 64, 705-727. Wahr, J. M., 1982. The effects of the atmosphere and oceans on the Earth’s wobble-I. Theory, Geophys. J. R. astr. SOC., 70, 349-372. Wahr, J. M., 1986. Geophysical aspects of polar motion, variations in the length of day, and the luni-solar nutations, in Space Geodesy and Geodynamics, pp. 281-313, eds Anderson, A. J. & Cazenave, A., Academic Press, Orlando. Woolard, E. W., 1953. Theory of the rotation of the Earth around its center of mass, Astr. Pap. A m . Ephern. Naut. Alman., vol. 15, part 1, US Government Printing Office, Washington, DC.
© Copyright 2026 Paperzz